8.9k citations · 11.5k across the 56 of their papers we have counts for
25 papers · 1 filter
How LLMs Detect and Correct Their Own Errors: The Role of Internal Confidence Signals
Dharshan Kumaran, Viorica Patraucean, Simon Osindero +2
Large language models can detect their own errors and sometimes correct them without external feedback, but the underlying mechanisms remain unknown. We investigate this through th…
Causal Evidence that Language Models use Confidence to Drive Behavior
Dharshan Kumaran, Nathaniel Daw, Simon Osindero +2
Metacognition -- assessing the quality of one's own cognitive performance -- guides adaptive behavior across species. Substantial research demonstrates that confidence signals can…
Perplexity Cannot Always Tell Right from Wrong
Petar Veličković, Federico Barbero, Christos Perivolaropoulos +2
Perplexity -- a function measuring a model's overall level of "surprise" when encountering a particular output -- has gained significant traction in recent years, both as a loss fu…
How Overconfidence in Initial Choices and Underconfidence Under Criticism Modulate Change of Mind in Large Language Models
Dharshan Kumaran, Stephen M Fleming, Larisa Markeeva +8
Large language models (LLMs) exhibit strikingly conflicting behaviors: they can appear steadfastly overconfident in their initial answers whilst at the same time being prone to exc…
Wanting to be Understood
Chrisantha Fernando, Dylan Banarse, Simon Osindero
This paper explores an intrinsic motivation for mutual awareness, hypothesizing that humans possess a fundamental drive to understand and to be understood even in the absence of ex…
Neural Compression of Atmospheric States
Piotr Mirowski, David Warde-Farley, Mihaela Rosca +7
Atmospheric states derived from reanalysis comprise a substantial portion of weather and climate simulation outputs. Many stakeholders -- such as researchers, policy makers, and in…
Genie: Generative Interactive Environments
Jake Bruce, Michael Dennis, Ashley Edwards +22
We introduce Genie, the first generative interactive environment trained in an unsupervised manner from unlabelled Internet videos. The model can be prompted to generate an endless…
Retrieval-Augmented Reinforcement Learning
Anirudh Goyal, Abram L. Friesen, Andrea Banino +13
Most deep reinforcement learning (RL) algorithms distill experience into parametric behavior policies or value functions via gradient updates. While effective, this approach has se…
Model-Value Inconsistency as a Signal for Epistemic Uncertainty
Angelos Filos, Eszter Vértes, Zita Marinho +7
Using a model of the environment and a value function, an agent can construct many estimates of a state's value, by unrolling the model for different lengths and bootstrapping with…
Top-KAST: Top-K Always Sparse Training
Siddhant M. Jayakumar, Razvan Pascanu, Jack W. Rae +2
Sparse neural networks are becoming increasingly important as the field seeks to improve the performance of existing models by scaling them up, while simultaneously trying to reduc…
Small Data, Big Decisions: Model Selection in the Small-Data Regime
Jorg Bornschein, Francesco Visin, Simon Osindero
Highly overparametrized neural networks can display curiously strong generalization performance - a phenomenon that has recently garnered a wealth of theoretical and empirical rese…
AlgebraNets
Jordan Hoffmann, Simon Schmitt, Simon Osindero +2
Neural networks have historically been built layerwise from the set of functions in , i.e. with activations and weights/parameters represented…
A Practical Sparse Approximation for Real Time Recurrent Learning
Jacob Menick, Erich Elsen, Utku Evci +3
Current methods for training recurrent neural networks are based on backpropagation through time, which requires storing a complete history of network states, and prohibits updatin…
A Deep Neural Network's Loss Surface Contains Every Low-dimensional Pattern
Wojciech Marian Czarnecki, Simon Osindero, Razvan Pascanu +1
The work "Loss Landscape Sightseeing with Multi-Point Optimization" (Skorokhodov and Burtsev, 2019) demonstrated that one can empirically find arbitrary 2D binary patterns inside l…
Adapting Behaviour for Learning Progress
Tom Schaul, Diana Borsa, David Ding +4
Determining what experience to generate to best facilitate learning (i.e. exploration) is one of the distinguishing features and open challenges in reinforcement learning. The adve…
Meta-learning of Sequential Strategies
Pedro A. Ortega, Jane X. Wang, Mark Rowland +21
In this report we review memory-based meta-learning as a tool for building sample-efficient strategies that learn from past experience to adapt to any task within a target class. O…
Distilling Policy Distillation
Wojciech Marian Czarnecki, Razvan Pascanu, Simon Osindero +3
The transfer of knowledge from one policy to another is an important tool in Deep Reinforcement Learning. This process, referred to as distillation, has been used to great success,…
Meta-Learning with Latent Embedding Optimization
Andrei A. Rusu, Dushyant Rao, Jakub Sygnowski +4
Gradient-based meta-learning techniques are both widely applicable and proficient at solving challenging few-shot learning and fast adaptation problems. However, they have practica…
Mix&Match - Agent Curricula for Reinforcement Learning
Wojciech Marian Czarnecki, Siddhant M. Jayakumar, Max Jaderberg +5
We introduce Mix&Match (M&M) - a training framework designed to facilitate rapid and effective learning in RL agents, especially those that would be too slow or too challenging to…
Kickstarting Deep Reinforcement Learning
Simon Schmitt, Jonathan J. Hudson, Augustin Zidek +8
We present a method for using previously-trained 'teacher' agents to kickstart the training of a new 'student' agent. To this end, we leverage ideas from policy distillation and po…
Population Based Training of Neural Networks
Max Jaderberg, Valentin Dalibard, Simon Osindero +9
Neural networks dominate the modern machine learning landscape, but their training and success still suffer from sensitivity to empirical choices of hyperparameters such as model a…
Sobolev Training for Neural Networks
Wojciech Marian Czarnecki, Simon Osindero, Max Jaderberg +2
At the heart of deep learning we aim to use neural networks as function approximators - training them to produce outputs from inputs in emulation of a ground truth function or data…
Understanding Synthetic Gradients and Decoupled Neural Interfaces
Wojciech Marian Czarnecki, Grzegorz Świrszcz, Max Jaderberg +3
When training neural networks, the use of Synthetic Gradients (SG) allows layers or modules to be trained without update locking - without waiting for a true error gradient to be b…
Decoupled Neural Interfaces using Synthetic Gradients
Max Jaderberg, Wojciech Marian Czarnecki, Simon Osindero +4
Training directed neural networks typically requires forward-propagating data through a computation graph, followed by backpropagating error signal, to produce weight updates. All…